Cutting Edge: TCR Ligation Triggers Digital Activation of NF-κB

Nuclear factor κB is a transcription factor of central importance in T cell biology, inducing expression of genes involved in proliferation, effector differentiation, and survival. Ligation of the TCR and CD28 initiates a complex series of cytosolic phosphorylation and ubiquitination events that culminate in the activation of NF-κB (Supplemental Fig. 1) (1).

Signal transduction pathways can be broadly characterized as digital or analog. Digital signaling is switch-like, specifying either on or off outcomes, with no intermediate state. Thus, once the threshold for signal activation has been met, the intensity of the output signal is constant. In contrast, analog signaling entails a graded response, with the intensity of the output signal proportional to the intensity of the input stimulus. Frequently, biological signaling cascades combine digital and analog features to achieve exquisite control over outcomes. For example, epidermal growth factor (EGF) activation of ERK combines digital and analog signaling mechanisms (2). Specifically, plasma membrane Ras nanoclusters activate ERK in a digital manner, but the number of signaling Ras nanoclusters is dependent on the strength of the input signal (EGF concentration). In this way, the number of on Ras nanoclusters tunes the magnitude of the output response, allowing the cell to appropriately respond to different concentrations of EGF.

The TCR-triggered ERK pathway in T cells also combines both digital and analog elements. One model suggests that although ERK phosphorylation is digital, the inhibitory activity of the Src homology 2 domain containing protein tyrosine phosphatase-1 (SHP-1) phosphatase is analog in nature (3). This model predicts that signals generated from weak TCR ligands are not strong enough to overcome SHP-1 inhibition and therefore fail to trigger the phosphorylation of ERK. In contrast, strong TCR ligands quickly overcome SHP-1 inhibition, yielding a defined amount of ERK phosphorylation and T cell activation (3, 4). Another model proposes that weak Ag signals through the TCR trigger activation of Ras-GRP (but not son of sevenless), yielding (weak) analog activation of Ras. In contrast, strong Ag signals trigger activation of both Ras-GRP and son of sevenless, resulting in digital activation of the Ras-ERK pathway (5). Although the mechanistic details of these models differ, both postulate that analog Ag signals through the TCR are converted to digital outputs at an early (receptor-proximal) point in the ERK signaling cascade.

In contrast to TCR activation of ERK, recent data suggest that stromal cell-derived factor-1 signaling through CXCR4 triggers ERK activation in T cells in an analog manner (6). Additionally, bypassing the TCR via PMA stimulation also results in analog activation of ERK (5). Thus, within a single cell, the same signaling module may be activated in either an analog or digital manner, depending on the nature of the stimulus.

Although there are now substantial data defining the TCR-to-ERK cascade as a digital signaling process (3–6), we are aware of no published reports examining the question of whether TCR activation of NF-κB occurs in a digital or analog manner. We thus conducted experiments to assess T cell NF-κB activation on a per-cell basis. These analyses reveal that TCR activation of NF-κB is digital in nature (i.e., with increasing TCR stimulation, the number of cells triggering the NF-κB pathway increases), whereas the magnitude of the NF-κB activation response is remarkably constant. All examined NF-κB signaling events exhibited digital behavior, including steps downstream of IκB kinase (IKK) activation (IκBα phosphorylation and degradation, RelA phosphorylation, and nuclear translocation) and upstream of IKK activation [Bcl10/Malt1 aggregation, protein kinase C (PKC) θ immunological synapse (IS) translocation]. Thus, analogous to the ERK signaling cascade, digitization of TCR signaling to NF-κB apparently occurs at a very early step in signal transmission.

Abs and stains used in this study were purchased from eBioscience (San Diego, CA): anti-human CD3 (OKT3) and PE anti-mouse CD4 (GK1.5); from Cell Signaling Technology (Beverly, MA): rabbit anti–phospho-IκBα (Ser32; 14D4), rabbit anti–phospho-RelA (Ser536; 93H1), and DRAQ5; from BD Biosciences (San Jose, CA): PE anti-IκBα (clone 25), PerCP anti-CD8 (53-6.7), and anti-human CD28 (CD28.2); from Invitrogen (Carlsbad, CA): Alexa 488-, Alexa 555-, and Alexa 647-goat anti-rabbit IgG; from Jackson ImmunoResearch Laboratories (West Grove, PA): goat anti-mouse IgG; from Santa Cruz Biotechnology (Santa Cruz, CA): rabbit anti-p65/RelA Ab (C-20); or purified from hybridoma supernatants: anti-mouse CD3 (145-2C11) and anti-mouse TCRβ (H57-597).

C57BL/6 (WT) mice, 4–8 wk old, were purchased from the National Cancer Institute (Frederick, MD). OTII mice (7) between 6 and 10 wk old were bred in-house. All experiments were approved by the Uniformed Services University of the Health Sciences Institutional Animal Care and Use Committee.

Jurkat T cells (clone E6-1, American Type Culture Collection, Manassas, VA) and the Carma1- and IKKγ-deficient mutants of Jurkat (8, 9) (kind gifts from Xin Lin and Shao-Cong Sun, respectively [University of Texas MD Anderson Cancer Center, Houston, TX]) were cultured at a density of 1–5 × 10⁵ cells/ml in RPMI 1640 plus 10% FBS. Following 16 h starvation in RPMI 1640 plus 0.2% FBS, cells were incubated on ice for 30 min with the indicated concentrations of anti-CD3 and anti-CD28, followed by 15 min incubation with 10 μg/ml anti-mouse IgG. Cells were activated by incubation in a 37°C water bath for 20 min. For PMA stimulation, Jurkat T cells were incubated on ice for 30 min with the indicated concentrations of PMA and 1 μM ionomycin. Cells were activated by incubation in a 37°C water bath for 25 min.

Construction and maintenance of D10 T cell lines expressing PKCθ-GFP or Bcl10-CFP+Malt1-YFP has been previously described (10, 11). D10 T cells were stimulated either with immobilized anti-CD3, PMA plus ionomycin, or with CH12 B cells loaded overnight with conalbumin, as indicated in the figure legends. For primary T cell stimulations, lymph nodes were harvested, and T cells were purified and stimulated with anti-TCRβ or OVA as previously described (12).

Jurkat T cells were fixed with 1.5% paraformaldehyde and permeabilized with BD Phosflow Perm Buffer II chilled to −20°C (BD Biosciences). Cells were blocked for 1 h with 1.25% normal mouse serum, then incubated overnight with PE–anti-IκBα. Primary T cells were fixed with 2% paraformaldehyde and stained with anti-CD8 and anti-CD4, followed by two washes, permeabilization, and overnight staining with Abs specific for phospho-IκBα or phospho-RelA. Secondary detection was with Alexa 647– or Alexa 488–anti-rabbit IgG. Flow cytometry was performed on a BD LSRII (BD Biosciences), and data analysis was performed using WinList 5.0 and FlowJo (Tree Star, Ashland, OR).

Stimulated D10 cells were fixed, stained, and imaged as described (10–12) using rabbit anti-RelA followed by Alexa 555–anti-rabbit and DRAQ5 (to label nuclei). The ratio of nuclear/cytoplasmic RelA was calculated as described (12).

To assess NF-κB activation on a per-cell basis, we employed a flow cytometry assay of IκBα degradation, as recently described (13, 14). Stimulation of Jurkat T cells with 5 μg/ml anti-CD3 resulted in a decrease in IκBα protein compared with unstimulated controls, indicating NF-κB activation (Fig. 1A). Notably, IκBα degradation occurred in most or all cells, and the amount of degradation appeared quite uniform. As expected, NF-κB essential modulator and Carma1-deficient Jurkat lines showed no IκBα degradation in response to anti-CD3 (8, 15), further validating the specificity of this assay.

We next performed an experiment to compare IκBα degradation in response to a broad range of anti-CD3 concentrations. As shown in Fig. 1B, across at least a 10,000-fold concentration range, two populations of cells are always apparent: a stimulated population with lower IκBα levels (dashed line) and an unstimulated population with higher IκBα levels (dotted line). Only the proportion of responding cells (but not the intensity of the degradation peak) changed as the anti-CD3 stimulus was increased. These data suggest that the strength of the anti-CD3 stimulus does not affect the magnitude of signaling, but rather the number of responding cells, consistent with an on/off (digital) pattern of activation.

We also performed an experiment stimulating the Jurkat cells with varying concentrations of PMA plus 1 μM ionomycin. In contrast to the results with anti-CD3, all concentrations of PMA yielded a single peak of IκBα fluorescence, which gradually decreased in intensity as the PMA concentration was increased (Fig. 1C). Intermediate values of IκBα fluorescence were most apparent in the 400–800 pg/ml concentration range. Thus, analogous to previously reported data regarding PMA-mediated activation of ERK (5), the PMA-stimulated degradation of IκBα is graded (analog). Taken together, the data in Fig. 1 suggest that IκBα degradation in response to CD3 ligation is digital, whereas PMA-stimulated IκBα degradation is analog.

The immediate downstream consequence of IκBα degradation is the nuclear translocation of NF-κB. To investigate the kinetics of nuclear translocation of RelA-containing NF-κB complexes, D10 T cells were stimulated with concentrations of anti-CD3 empirically determined to yield high, moderate, and low percentages of cells with NF-κB nuclear translocation. As shown in Fig. 2A, the kinetics of RelA nuclear translocation were similar across all stimulation conditions. The percentage of cells showing nuclear RelA sharply increased by 0.5 h poststimulation and remained relatively unchanged through at least 3 h poststimulation. Between 3 and 4 h poststimulation, the percentage of cells with nuclear RelA markedly declined, approaching basal levels by 6 h poststimulation. Thus, higher concentrations of anti-CD3 recruited a greater proportion of responding cells, but the kinetics of the response were essentially identical across concentrations.

Next, the average anti-RelA fluorescence intensity in the cytosol and nucleus was quantified for cells at 0, 0.5, and 4 h poststimulation, and the per-cell ratio of nuclear to cytoplasmic RelA was used as a measure of the degree of NF-κB translocation. At both 0.5 and 4 h post anti-CD3 stimulation (Fig. 2B, Supplemental Fig. 2A), the proportion of RelA nuclear translocation per cell was constant across all stimulation concentrations. In contrast, nuclear translocation of RelA increased as the concentration of PMA was increased (Fig. 2B, Supplemental Fig. 2B). Thus, consistent with the results in Jurkat T cells, stimulation of D10 T cells with increasing concentrations of anti-CD3 did not affect the magnitude of the output signal (RelA nuclear translocation), whereas increasing PMA concentrations clearly increased the output intensity. Taken together, these data indicate that NF-κB nuclear translocation occurs in a digital fashion in response to signals through the TCR, but in an analog manner in response to PMA.

IκBα degradation and NF-κB nuclear translocation are terminal steps in NF-κB activation, which occur downstream of IKK activation. To determine whether digitization of the NF-κB cascade occurs at the level of or upstream of the IKK complex, we examined signaling events known to occur upstream of IKK activation. Punctate and oligomeric killing or activating domains transducing signals (POLKADOTS) are cytoplasmic foci containing Bcl10, Malt1, and a number of additional NF-κB signaling molecules. These structures form post TCR stimulation in a manner highly correlated with successful signal transmission by Bcl10 and Malt1 (10, 11, 16). The concentration of stimulatory anti-CD3 Ab affected the percentage of responding cells, as higher anti-CD3 Ab concentrations resulted in an increased percentage of cells displaying POLKADOTS and nuclear RelA (Supplemental Fig. 3A, 3B). Indeed, at all tested concentrations of anti-CD3, 100% of the cells that formed POLKADOTS also had RelA nuclear translocation, and the distribution of RelA nuclear/cytoplasmic intensities was remarkably similar at all three concentrations (Fig. 2C, Supplemental Fig. 3C).

In contrast, in response to PMA stimulation, the distribution of nuclear/cytoplasmic RelA intensities increased in proportion to the PMA dose. Moreover, at the 1 ng/ml and 0.1 ng/ml PMA concentrations, substantial numbers of cells with POLKADOTS had background levels of RelA translocation (Fig. 2C). Thus, in these cells, PMA stimulation triggered POLKADOTS formation, but failed to trigger RelA nuclear translocation. These observations are consistent with termination of the NF-κB signaling cascade downstream of POLKADOTS formation and upstream of RelA activation. Thus, unlike stimulation with anti-CD3, stimulation with suboptimal levels of PMA yields an intensity of downstream signaling that is proportional in magnitude to PMA concentration and is abortive in a subset of cells. Together, these data suggest that signal digitization in response to anti-CD3 stimulation occurs at the level of or upstream of Bcl10/Malt1 activation.

To examine whether NF-κB activation events upstream of Bcl10 are digital or analog, we quantified the percentage of D10 T cells showing translocation of PKCθ to the IS. Our previous data showed a high degree of correlation between cells with POLKADOTS and cells with PKCθ translocation to the IS (10). Thus, given the data in Fig. 2C and Supplemental Fig. 3, it was not surprising that cells with RelA nuclear translocation invariably showed PKCθ translocation to the IS (Fig. 2D). Although the percentage of T cell/B cell conjugates declined with Ag dose, no apparent difference in the efficiency of PKCθ translocation to the IS was observed (Supplemental Fig. 4). Combined with our previous observation that the percentage of cells with PKCθ translocation and POLKADOTS formation also declines with Ag dose (10), we conclude that the strength of Ag signaling through the TCR influences the probability that the TCR will be activated, but not the strength of signaling in the PKCθ-to-NF-κB cascade.

Taken together, the above data suggest that digitization of NF-κB activation occurs well upstream of IKK activation, at the level of or upstream of PKCθ activation (Supplemental Fig. 1). Thus, digitization of the NF-κB signal is a receptor-proximal event, occurring at a step early in signal transmission.

To determine if the results from Jurkat and D10 T cell lines could be generalized to the activation of primary T cells, C57BL/6 T cells were stimulated with increasing concentrations of anti-TCRβ Ab. We monitored phosphorylation of IκBα by flow cytometry, as previously described (12). Stimulation with 4–100 μg/ml anti-TCRβ resulted in the appearance of a phosphorylation peak in both CD4⁺ and CD8⁺ T cells. Consistent with the behavior of Jurkat T cells (Fig. 1B), as the stimulatory Ab concentration increased, both CD4⁺ and CD8⁺ T cells showed an increase in the percentage of cells with IκBα phosphorylation. However, there was no change in the per-cell intensity of IκBα phosphorylation (in the responding peak) across anti-TCRβ concentrations (Fig. 3A). Analogous results were obtained when we examined the activation-associated phosphorylation of RelA at S536 (17). Again, the percentage of cells with RelA phosphorylation increased as the concentration of anti-TCRβ Ab increased, but the per-cell intensity of phospho-RelA in the responding population was constant. Taken together, the data from Fig. 3A and 3B indicate that once the threshold for TCR signal transmission has been met, IκBα and RelA phosphorylation occurs in a digital manner in primary CD4⁺ and CD8⁺ T cells.

Ab-mediated cross-linking of the TCR is a potent method of stimulating T cells, which may result in artificially high levels of signaling and confound the analysis of signaling pathways. Therefore, we used CD4⁺ T cells from OTII transgenic mice to examine the phosphorylation of IκBα in response to Ag. OTII CD4⁺ T cells stimulated with null peptide or 0.04 μg/ml OVA peptide showed no detectable IκBα phosphorylation, whereas 0.2–5 μg/ml OVA peptide yielded a population of cells with phosphorylation of IκBα (Fig. 3C). Consistent with data from Jurkat cells (Fig. 1), D10 cells (Fig. 2), and anti-TCR–stimulated primary T cells (Fig. 3A, 3B), the percentage of responding cells increased with increasing peptide concentration, but the per-cell level of IκBα phosphorylation was constant across Ag doses. Together, the data in Fig. 3 suggest that TCR-mediated NF-κB activation in primary CD4⁺ and CD8⁺ T cells is digital in nature. Furthermore, digital activation occurs whether the activating stimulus is an anti-TCR Ab or an Ag-loaded APC.

Overall, our data indicate that TCR signaling results in digital activation of NF-κB in T cell lines and primary T cells in response to either Ab cross-linking of the TCR or antigenic stimulation. Differences in Ag availability control the number of cells successfully activated rather than the magnitude of activation in individual cells. Signal digitization apparently occurs at the level of or upstream of PKCθ. Thus, as in ERK signaling from the TCR (3–6), signal digitization to NF-κB occurs at a very early step in the signaling cascade.

This work, showing digital TCR activation of NF-κB, in combination with previous work documenting digital TCR-to-ERK signaling, provides strong evidence that the TCR has evolved to function as a digital sensor of peptide Ag. The particular tuning of this digital machine guarantees that strongly activating foreign Ags, but not weakly activating self-Ags, generate productive signals to downstream transcription factors (3). Furthermore, once the threshold for signal transmission has been met, digital signaling through the TCR is likely to ensure that the responding cell activates the precise levels of gene transcription required to successfully complete complex proliferation and effector differentiation programs. Although current models of TCR triggering of the Ras-ERK pathway (3, 5) do not explicitly include a mechanism for initiation of NF-κB signaling, it seems logical that the mechanisms of digital activation of ERK and NF-κB signaling would be identical or mechanistically linked to prevent the circumstances under which ERK, but not NF-κB (or vice versa), is activated. The elucidation of early steps that specify digital TCR signaling is thus clearly an important area for further investigation.

Importantly, while this manuscript was in revision, another group reported digital NF-κB activation (with certain analog components) downstream of TNF stimulation (18). It is thus likely that digital activation of NF-κB can be generalized to many distinct receptor-triggered NF-κB signaling cascades.

We thank Xin Lin and Shao-Cong Sun for cell lines.

Disclosures The authors have no financial conflicts of interest.